Conversation
Highlights

How Collate Turned Open Source into a GTM Engine: From Zero to 12,000+ Community Members

Most infrastructure founders ask the wrong question about open source. They debate whether to open source their code. The real question is whether you can architect your open source project to systematically generate commercial demand.

Suresh Srinivas had built Hadoop’s core at Yahoo, scaled Hortonworks, and architected Uber’s data systems processing 4 trillion events daily. When he founded Collate in 2021, he made a contrarian bet: build a metadata platform from scratch as open source, invest two years purely in community, then watch inbound sales follow.

In a recent episode of BUILDERS, Suresh, Co-Founder and CEO of Collate, broke down how this strategy generated 12,000+ community members, 3,000+ production deployments, and a commercial pipeline that’s nearly 100% inbound.

The Metadata Problem Nobody Was Solving

At Hortonworks, Suresh watched a pattern repeat across customers. Companies had powerful data platforms—Hadoop clusters storing and processing massive datasets—but struggled to extract business value. “We had built all these platforms, amazing platforms that stored lots of your data, crunched lots of your data,” Suresh said. “But what I saw in Hortonworks is people were struggling to find value from it and build business value from it.”

He joined Uber in 2018 to understand the user side. Even at a world-class data organization managing 4 trillion daily events, where “every ride and the driver match, everything was happening through data,” the same problems existed. Data practitioners couldn’t find the right datasets, didn’t understand lineage or business context, and made mistakes that triggered GDPR violations.

The diagnosis? “Many of the data problems can be solved if you have end to end full context of metadata,” Suresh explained. But existing tools fragmented metadata across discovery, observability, and governance—none built with unified vocabularies or clean data models.

Why Building From Scratch Beat Spinning Out

Uber had internal metadata tooling that could have been spun out—the standard playbook for infrastructure startups. Suresh chose differently.

“We took actually a different approach. We ended up building the open source project from the ground up after leaving Uber,” he said. The reason was architectural. “When you look at Uber technology, it is built for very large scale. But the solution that we have built is useful for small, medium, large companies.”

This wasn’t about avoiding complexity for its own sake. It was about GTM expansion strategy. Building only for Uber-scale would have locked Collate into enterprise-only sales with 18-month cycles. Building for all company sizes from day one opened the market while maintaining the sophistication that enterprises need.

The tradeoff? Zero brand recognition. They would need to prove themselves through execution, not pedigree.

The First 90 Days: Code, Not Pitches

While most founders spend their first 90 days pitching customers, Collate’s first 90 days looked different. “We didn’t even have a startup,” Suresh recalled. “Our first 90 days was very interesting. So we decided to build an open source project called Open Metadata. I was sitting and coding with my co-founder and then later went and raised money.”

They open sourced in August 2021 and began shipping at high velocity. The product architecture was deliberate: unify data discovery, observability, and governance into one platform with a clean metadata vocabulary and unified data model.

This architectural decision created natural commercial conversion paths. By solving a complete workflow—not a point solution—they made it obvious where managed infrastructure and enterprise features would add value.

100+ Daily Conversations: The Feedback Loop That Scales Velocity

What emerged next was something most product teams can only dream about. “We have around 100 plus conversations that we have with our open source community contributors, participants,” Suresh said. “It’s a product manager’s dream. You get to interact with your customers on a daily basis.”

These conversations happened in context—users deploying across Snowflake, Databricks, and other integrations, hitting real-world edge cases. “Every release that goes out and we operate in a complex data ecosystem environment, we get immediate feedback on what is working, what is not working. With Snowflake, this problem happened with Databricks, some other problem happened.”

The result? Validated product decisions at a pace closed-source competitors couldn’t match. No guessing what features matter. No waiting for quarterly business reviews to learn what’s broken.

From Feedback to Code: Scaling Contribution

Four years in, Collate has 400+ code contributors. That’s not just feature requests or bug reports—it’s engineers submitting pull requests that ship in production.

“If you do the community the right way, and encourage them and review their PRs, the pull request that they’re sending to you and not ignorant, you can actually increase the number of contributors and increase your product velocity,” Suresh explained.

This wasn’t passive community management. It required active curation—reviewing PRs promptly, incorporating feedback into the roadmap, giving contributors real influence over project direction. “That collaboration is very important,” Suresh emphasized.

The multiplier effect is real. 400 contributors shipping improvements across a complex data ecosystem is development velocity no startup team could self-fund.

The Inbound Sales Engine

Two years into community building, commercial traction followed. Their first major customer was a top Portuguese bank. “It’s very hard to sell to banks if you know about the data landscape,” Suresh noted. They won because the bank’s technical team had already validated the technology.

Today, “most of our sales is through inbound and that happened because our go to market strategy was centered around open source.” With 12,000+ community members and 3,000+ deployments, nearly every commercial conversation starts with practitioners who’ve used Open Metadata in production.

The conversion logic is straightforward. “People use our open source project and then they want to use the world class experts. Instead of they having to actually support and build and take care of the infrastructure and the deployment, they want somebody who are experts to do that for them,” Suresh explained.

Add differentiated features built on controlled cloud infrastructure, and you have an open core model where usage naturally reveals willingness to pay.

Timing the CMO Hire

As commercial traction grew, Collate faced a messaging challenge. “Our initial audience is technical folks and we, both the founders and the founding team are made of technical people who have seen this problem, who had this problem, who solved this problem. We can speak the language of the technical folks and communicate to them and connect with them.”

But enterprise deals require articulating business value to executives who care about compliance risk, AI readiness, and ROI on data initiatives—not metadata schemas.

Their solution: hire a CMO with deep data domain expertise after establishing product-market fit. “As we grow, we also need to talk to the business leaders, communicate the problems at the business level. The business impacted it creates the problem it solves,” Suresh said.

The sequencing mattered. Hiring for executive messaging before technical adoption was proven would have diluted their credibility with practitioners—the users actually driving adoption.

Playing the Long Game With Builder-Culture Companies

Suresh’s experience selling to companies like Uber offers a lesson in patience. Internal tools at tech giants often start ahead but become technical debt as teams move to new problems. “The internal teams, they build it and they move on to something else,” he observed.

Meanwhile, focused startups close the gap. “There’s a startup that might be farther behind compared to in house tools built at technology companies like Uber. But they are sharply focused on the problem that they’re trying to solve. That means that within a two or three years they not only catch up with the in house tools that these large companies build, they will overtake it.”

His advice? “Keep in touch with these larger companies. Your technology will improve and you will have better conversation with larger technical companies.” The buying window opens when maintenance burden exceeds building pride—typically 24-36 months post-internal launch.

The Semantic Intelligence Bet

Looking ahead, Suresh sees metadata as foundational infrastructure for reliable AI. “AI with the semantics enabled for data, where AI can understand the data is really necessary to unleash AI on your data and get much better value,” he said.

The thesis: AI won’t ask clarifying questions about what “revenue” means in your organization. It won’t have hallway conversations to build context. “Semantics through ontological technologies is becoming very important,” Suresh explained.

But the endgame isn’t technology for its own sake. “It’s not about technology, it’s about creating business value,” Suresh emphasized.

What This Means for Infrastructure Founders

Collate’s journey validates a specific approach to open source GTM:

Architect for commercial conversion from day one. Don’t just open source code—design a platform that solves complete workflows and reveals natural upgrade paths.

Invest in community before sales infrastructure. Two years of pure community building generated an inbound engine that now drives nearly 100% of deals.

High-velocity releases build credibility faster than brand. Without Yahoo or Uber’s halo, shipping cadence and responsive PR reviews proved commitment to technical buyers.

Time your go-upmarket hiring deliberately. Technical founders drove practitioner adoption. Domain-expert CMO translated value for executives. The sequencing prevented diluting credibility during the critical adoption phase.

For founders building technical infrastructure, the lesson is clear: the longest path to market—community before commerce—can create the most defensible GTM advantage.